Tilburg University

Are international deposits tax-driven?

Huizinga, Harry; Nicodeme, G.

Published in:Journal of Public Economics

Publication date:2004

Link to publication

Citation for published version (APA):Huizinga, H. P., & Nicodeme, G. (2004). Are international deposits tax-driven? Journal of Public Economics,88(6), 1093-1118.

General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

- Users may download and print one copy of any publication from the public portal for the purpose of private study or research - You may not further distribute the material or use it for any profit-making activity or commercial gain - You may freely distribute the URL identifying the publication in the public portal

Take down policyIf you believe that this document breaches copyright, please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.

Download date: 03. Feb. 2019

1

Are International Deposits Tax-Driven ?

Harry Huizinga* European Commission, Tilburg University, and CEPR

and

Gaëtan Nicodème*

European Commission, and Solvay Business School (ULB)

Final version: March 2003

Abstract: This paper investigates the impact of tax policy on international depositing. Non-bank international deposits are shown to be positively related to interest income taxes and to the presence of domestic bank interest reporting. This suggests that international deposits are in part intended to facilitate tax evasion. At present, only part of international interest flow are covered by either non-resident interest withholding taxes or international exchange of information. This incomplete coverage may be a reason that these policies currently appear to have little impact on international depositing.

* We thank participants in the conference on "World Tax Competition" organized by the Office of Tax Policy Research of the University of Michigan Business School and the Institute for Fiscal Studies of May 2001 in London, and especially Julian Alworth and Andreas Haufler for helpful comments. Helpful comments from Giuseppe Carone, Joel Slemrod, Guttorm Schjelderup and two anonymous referees are also gratefully acknowledged. The findings, interpretations and conclusions expressed in this paper are entirely those of the authors. They should not be attributed to the European Commission.

2

1 . Introduction

Countries typically tax the worldwide interest income of their residents. By now,

the tax authorities in most OECD countries require domestic banks to report interest

payments to domestic residents. In contrast, no comprehensive system of international

exchange of bank interest information exists to date. This, combined with the generally

low taxation of international bank interest at source, implies that the international

recipient of bank interest can evade all taxation of this income with relative ease.

In the minds of European policy makers, this has been a serious problem since at

least the 1980s, as evidenced by the introduction in 1989 of a first proposal for a

European directive towards a common minimum withholding tax on interest. In 1998, a

second proposal for a directive was published that gave EU member states the option to

tax interest accruing to non-residents at source or to exchange information with other

countries. In November 2001, EU finance ministers abandoned the idea of co-existing

withholding taxes and information exchange, and instead stated their intention to move

towards generalized information exchange by 2010. Until then, several countries, namely

Austria, Belgium and Luxembourg, will be free to levy a minimum withholding tax

instead, with the understanding that 75 percent of the tax revenues are passed on to the

residence-country tax authorities. These intended policies have been laid down in a

proposal for a directive in July 20011.

The adoption of a directive in the area of international interest taxation would be

the first major international agreement in the area of capital income taxation, or for that

matter of direct taxation in general. The further development of policy in this area (to

include, say, countries outside the EU, or to extend coverage to dividends) is hampered

by a lack of empirical analysis of international interest tax evasion. A main impediment

to research in this area has been the limited data on the international ownership of bank

deposits and other financial assets. Countries are presumably restricting access to these

1 See European Commission (2001). A stated condition for the adoption of this directive is that

the European Union reaches agreement with several third countries, notably Switzerland, to institute similar anti-evasion measures in these countries.

3

data to protect the employment and profits of their domestic banking sectors.2 More

discussion at the international level of the potential roles of banks in tax evasion and

money laundering schemes may some day force more openness, but for now data on

bilateral banking flows remain confidential. Data of this kind, however, are collected by

the Bank for International Settlements (BIS), and have been made available for this study

on the condition that data on bilateral banking flows are not disclosed.

The main purpose of this paper is to see to what extent international banking

flows reflect tax policy and efforts to enforce it. Tax determinants first are the residence-

based interest income and wealth taxes that de jure typically apply to worldwide income

and wealth. To aid enforcement, many countries by now require their banks to report

interest payments to domestic residents to the tax authorities. To enable international

enforcement, banking countries in some instances also supply information to foreign tax

authorities. Data on both types of information provision have been collected for this

study. Finally, the analysis also takes into account that international interest payments

may be subject to an interest withholding tax in the source country.

Our empirical results suggest that interest income taxation has encouraged

international depositing, at least during the high-interest-rate period of the 1980s.

Domestic bank interest reporting also appears to contribute to international bank

placements. There is less evidence that interest withholding taxes discourage such

depositing, perhaps because non-resident withholding taxes are typically rather low and

imposed by relatively few countries. Similarly, there is little evidence that international

information exchange – for 1999 data – has a strong impact on bilateral depositing.

Again, a reason may be the haphazard pattern of international information exchange at

present. Truly generalized withholding taxes or information exchange in principle affect

the international depositing decision as much as domestic tax policy, and hence can be

expected to have a significant impact on international depositing patterns.

Several authors have previously examined the determinants of international

banking flows. Grilli (1989) relates non-bank and inter-bank deposits to interest and

2 Countries with relatively few internationally active banks may in addition see a need to retain information in order to maintain the confidentiality of bank-level information. Countries may originally

4

dividend taxes, capital flows, an index of bank secrecy, GNP, and a trend. He finds that

non-bank deposits are influenced by taxes on interest and by bank secrecy, while inter-

bank deposits are driven by the size of the source economy and by the taxation of

dividends (suggesting that bank accounts might be used to park money meant for later

financial transactions). Alworth and Andresen (1992) further estimate a gravity model to

explain the determinants of non-bank bilateral deposit flows using data up to 1990.3

These authors include several bank-system variables such as the (bilateral) difference in

reserve requirements, the bank-country interest withholding tax, and an index of its bank

secrecy. The withholding tax and bank secrecy variables, as part of interacted variables,

are shown to be determinants of cross-border deposits. More recently, Fornari and Levy

(2000) have estimated the determinants of bilateral cross-border deposit inflows for a

group of 6 industrialized countries. These authors place special emphasis on financial

structure variables such the stock market capitalization to GDP, stock market volatility

differences and the trading volume of the stock market.

As Alworth and Andresen (1992), the present paper examines the determinants of

bilateral international depositing with a focus on taxation. This paper differs, however, in

that we have somewhat more detailed information on the tax regime and the availability

of bank information to tax authorities. In particular, the present paper includes personal

interest income and wealth taxes and distinguishes between the domestic and

international availability of bank information to tax authorities.

Several theoretical papers have also examined tax policy towards mobile financial

capital. Janeba and Peters (1999), for instance, consider the issue of discrimination

against internationally mobile capital given that countries set tax rates non-cooperatively.

Huizinga and Nielsen (2000) show that an internationally agreed minimum withholding

have started to collect this information to monitor monetary developments rather than to check the competitive positions of their banking sectors. 3 Recently several papers have also applied the gravity approach to investigate capital flows other than cross-border deposits. Portes and Rey (1999), for instance, show that bilateral portfolio equity investments reflect variables proxying (private) information availability, such as international telephone calls and multinational bank branches. Along similar lines, Ahearne, Griever, and Warnock (2000) find that U.S. holdings of a country’s equities are positively related to the share of that country’s stock market that is listed on U.S. exchanges. This is attributed to the fact that a listing in the U.S. lowers information costs for U.S. investors.

5

tax on interest, that is only binding for a small country, can benefit all countries, if in fact

all countries are induced to increase their interest tax rates. Bacchetta and Espinosa

(1995) argue that it may be in a country’s own interest to provide information about bank

interest payments to non-residents, as this enables the interest-receiving country to

increase its own income tax rate. This in turn reduces the incentive for residents of the

information-providing country to place their savings abroad. In a repeated game

framework, Bacchetta and Espinoza (2000) further study the joint determination of taxes

on international investment income and information-exchange clauses in double taxation

treaties. They find that information exchange may be part of a (sustainable) tax treaty if

there is a reciprocity requirement, when there is a high cost of negotiation, or with one-

way capital flows. Also in a repeated game setting, Huizinga and Nielsen (2003) examine

countries’ exclusive choice between non-resident withholding taxes and information

exchange (as provided for by the European Commission’s draft directive of 1998, see

European Commission (1998)). Two countries choosing the same regime (either

withholding taxes or information exchange) and a mixed regime (one country choosing

withholding taxes and the other information exchange) are all possible equilibria of the

regime selection game. Information exchange performs relatively well, and is more likely

to be chosen in equilibrium, if governments apply a relatively low discount rate to future

outcomes. In the following, section 2 discusses the data used in this study. Section 3

presents the empirical results, and section 4 concludes.

2. The data

2.1 International deposits

The BIS has collected data on the external liabilities of reporting country banking

systems since 1983, and on external deposits from 1996 onwards4. The external liabilities

and deposits of BIS reporting countries for 1999 are reported in Table 1. These figures

represent all currencies. From the table, we see that the UK and the US have the largest

4 External deposits comprise all claims by non-residents on banks and bank-like reporting institutions with evidence of deposit not in the form of negotiable securities. Apart from external deposits, external liabilities include marketable instruments such as negotiable debt securities, bonds and short-term negotiable instruments, derivative instruments on-balance sheet, and working capital. See Bank for International Settlements (2000b).

6

external liabilities at � 1.8 trillion and � 1.0 trillion, respectively. Among the smaller

countries, the Cayman Islands and Switzerland have about � 0.6 trillion foreign liabilities,

while Luxembourg has around � 0.4 trillion. The total external liabilities of banks in the

BIS area amount to � 9.0 trillion. Total liabilities are divided between bank and non-bank

liabilities. Bank liabilities are debts to other banks, and non-bank liabilities are debts to

individuals, public institutions and to businesses.5 As seen in the second column, non-

bank liabilities are less than half of total liabilities in all reporting countries. For the BIS

area, non-bank liabilities stand at 24 percent of total liabilities. Interestingly, non-bank

liabilities are highest in Switzerland and the Cayman Islands at 48 and 42 percent of total

liabilities, respectively. External deposits are represented in the third column. External

deposits are shown to be the lion’s share of external liabilities.6 For the BIS area as a

whole, external deposits are 92 percent of external liabilities. The last column indicates

that non-bank external deposits are 25 percent of total external deposits.

It is also interesting to consider to what extent a country’s residents maintain

deposits abroad. To proceed, let dij be the non-bank deposits in country i owned by the

residents of country j (with i different from j). We can now define country j’s exports of

non-bank deposits (as part of capital exports) or Ej, and country i’s imports (as part of

capital imports) or Ii as follows,

Ej = �≠ ji

ijd Ii = �≠ij

ijd

To see how important these non-bank deposit exports and imports are, we can

relate them to the total Non-bank deposits in a country’s banking system and to the

worldwide ownership of non-bank deposits by a country’s residents. Specifically, let Di

be the total non-bank deposits in country i’s banking system. The worldwide ownership

of non-bank deposits by residents of country i then can be defined as Oi = Di + Ei – Ii .

The share of non-bank deposits owned by residents of country i held abroad is given by

5 These businesses include non-bank financial institutions such as mutual funds, hedge funds, and insurance companies. 6 Note that not all countries report separate data for external liabilities and deposits on a country basis. In the last several years, the rapid growth in external bank liabilities has resulted in a larger share of external bank liabilities in total external liabilities.

7

is = ii OE / . Net deposit imports cause a country’s banking system to be larger than it

would otherwise be. We can define the expansion ratio of a country’s banking system on

account of its net non-bank deposit imports as ig = iii OEI /)( − . This expansion is

measured relative to the hypothetical case where the banking system exactly

accommodates the non-bank deposits owned by the country’s residents. The expansion

ratio is a rough index of how much a particular banking system gains or loses on account

of its net non-bank deposit imports.

Table 2 provides data on aggregate deposit exports and imports and other derived

variables for 1998.7 Switzerland and the United Kingdom are shown to be net exporters

of deposits (bank and non-bank deposits together) from the first 2 columns, while they

are net importers of non-bank deposits from the 2 next columns. Net inflows of non-bank

deposits thus are more than off-set by net outflows of bank deposits. At any rate,

incoming non-bank deposits are recycled as outgoing bank deposits. Conversely, the

United States is a net exporter of non-bank deposits, and a net importer of bank deposits

(as net exports of non-bank deposits exceed net exports of overall deposits). Other net

exporters of non-bank deposits are Australia, France, Italy, Japan, Norway, and Spain.

Next, we turn to the share of non-bank deposits owned by residents held abroad.

Ireland leads here with 33 percent, reflecting its relatively high exports of non-bank

deposits. Australia, Canada, Denmark, Finland and Norway instead have foreign shares

of total non-bank deposit ownership at less than 5 percent, indicating relatively closed

banking systems. Finally, we consider the expansion rate of the banking system due to

net non-bank deposit imports. Switzerland is shown to be a large net non-bank deposit

importer, and correspondingly is calculated to have a banking expansion rate of 19%. The

United States and Spain, in contrast, display relatively large banking sector 'contractions'

on account of large net non-bank deposit exports. To increase the national coverage

somewhat, Table 3 provides information on exports and imports of bank liabilities rather

than bank deposits. Hong Kong registers as an additional net exporter of non-bank

liabilities, while the Bahamas is shown to be a strong net importer of non-bank liabilities.

7 We chose 1998 as the total non-bank banking system deposits published for 1999 by euro-area countries include shares in money market funds.

8

2.2 The tax system

Countries typically tax different types or income at different rates. Since 1983,

increasingly many countries have opted for dual tax systems with different tax rates for

earned and capital income. Capital income may again be taxed differently depending on

whether it takes the form of interest, dividends, or capital gains. In practice, even finer

gradations are found (especially with respect to international capital income flows) where

separate rates of tax are applied to bond interest, bank interest, or interest from a loan

secured by real estate. Wealth taxes tend to be less specific, although some countries

make distinctions between taxes on financial wealth (which could be divided into

portfolio wealth or business ownership), and real estate. Throughout, we have attempted

to identify the taxation of interest from deposits and wealth in the form of deposits as

regards individuals.

Table 4 provides the effective interest income and wealth taxes applied to bank

deposits in 1999 in most BIS reporting countries. Both taxes generally apply to

worldwide interest income and wealth, and take into account sub-national taxation of

interest in several cases, such as Canada and Denmark. In 1999, Austria, Belgium,

Finland, France, Greece, Ireland, Italy, Japan, Portugal, Sweden, and the United

Kingdom maintained dual (or multiple) income tax systems with a relatively low tax rate

for interest income. In most cases, the dual income tax system was introduced during the

1983-1999 period, with a view to discourage tax evasion and to lower compliance costs.

These introductions were probably at least in part meant to reduce the incentive to evade

the taxation of domestic capital income such as interest income8. Since 1983, the average

statutory interest income tax has declined gradually, as seen in Figure 1. Deposit interest

rates have declined as well, and hence the interest tax burden expressed as a percentage

of principal (and calculated as the statutory interest rate times the deposit interest rate)

has declined even more, as also seen in Figure 1.

8 Recent tax reforms continue the movement away from synthetic income tax systems. At the start of 2001, the Netherlands also introduced a dual system with a tax rate of 30 percent on a (deemed) return on capital income of 4 percent. This amounts to a wealth tax of 1.2 percent per annum to replace the previous wealth tax of 0.7 percent.

9

Table 4 also provides information about wealth taxes in place in 1999. These

annually assessed wealth taxes exclude taxes on intergenerational transfers such as estate

taxes. Since 1983, several countries have eliminated their regular wealth taxes (Austria’s

ended by 1994, Denmark’s by 1997, and Germany’s by 1997). France relinquished its

‘old’ wealth tax by 1986, to introduce a ‘new’ wealth tax in 1988. Overall, the average

wealth tax has declined significantly since 1983 (see Figure 2). Finally, we turn to non-

resident interest withholding taxes9. In 1999, only 4 countries, namely Australia, Japan,

Portugal, and Switzerland, levy positive withholding taxes on any outgoing bank interest

flows, as seen in the table. In several instances, interest paid by banks has been taxed at

lower non-resident withholding tax rates than other interest. The U.S., for instance, has

maintained a statutory exemption for bank interest throughout the period under

consideration, even though it levied a non-resident (non-treaty) interest withholding tax

of 30 applied to bond interest up to 1984. The U.K. similarly exempts bank interest on

bank claims with a maturity of less than a year including regular current account and

savings account deposits. Switzerland is a major financial center that continues to tax the

bank interest accruing to non-residents, even though this country has also reduced the

non-treaty tax rate of 35 percent to 12.5 percent or less in all but 5 cases10. Austria and

France are among the countries that have abolished non-resident withholding taxes in

1993 and 1997, respectively. Overall, the average statutory non-resident interest

withholding tax has declined since 1983, as seen in Figure 3. The withholding tax burden,

as a percentage of principal, has diminished even more, reflecting the decline in deposit

interest rates.

2.3 Access to bank information and international information exchange

Taxes on bank interest that are not withheld by the paying bank have to be

collected from the depositor. To make enforcement in this case realistic, the tax authority

9 See also Zee (1998) for an exposition of the role of withholding taxes in taxing international portfolio income. 10 In the case of Switzerland, many deposits are held in fiduciary accounts that de jure are inter-bank accounts not subject to withholding taxation, even if the ultimate beneficiaries are individuals.

10

needs to have independent access to bank information. Access to bank information for tax

purposes, either domestic or international, has been far from straightforward, as

documented in a comprehensive recent report by the OECD (2000).11 A first requirement

is that the banks themselves maintain the information that is required for tax enforcement

and that they do not open anonymous or numbered accounts. As indicated by OECD

(2000), the vast majority of OECD tax authorities can obtain bank information to combat

domestic tax evasion. Information provision – either domestic or international – can be

categorized as spontaneous (on the initiative of the information provider), on request, or

automatic. Tax authorities that request specific account information have to follow due

procedures – administrative or legal – to make the request. To make specific requests, tax

authorities need to already have some specific information on which to base the request.

Information provided on request is thus not likely to lead to across-the-board tax

enforcement.

This leaves the automatic and periodic provision of bank information as the only

viable way to enforce taxation. As seen in OECD (2000, Appendix 1), 15 OECD

countries require their banks to generally report ‘interest paid and to whom it is paid’.12

These countries were requested to indicate when they started to require their domestic

banks to automatically report interest payments to domestic residents. The answers

received are reflected in Table 5. As seen in the table, during the 1980s and early 1990s

several countries additionally required domestic interest reporting. By 1999 about two

thirds of the countries required automatic domestic information provisioning regarding

interest payments.

International automatic information exchange requires some international legal

agreement – in addition to domestic regulation. The legal basis can be a bilateral tax

treaty, which in many cases is modeled after the OECD Model Convention on Income

and Capital13. Article 26 of this convention requires contracting States to ‘exchange such

11 The OECD’s work to promote exchange of information, as reflected in this report, has been motivated by a drive against money laundering as much as by a desire to counteract tax evasion. 12 Frequently other information, for instance on account balances or on securities held in custody, is exchanged as well. 13 In April 2002, the OECD released a new model for (non-binding) bilateral agreements concerning the effective exchange of information in tax matters, following the work of the Global Forum Working

11

information as is necessary for carrying out the provisions of this Convention or of the

domestic laws of the Contracting States concerning taxes covered by the Convention

insofar as the taxation thereunder is not contrary to the Convention’. All OECD members

except Luxembourg and Switzerland can obtain bank information for the purpose of

exchange of information under tax treaties as set out in the Model Convention.14 Several

multilateral agreements that can serve as a basis for information exchange exist as well.

For instance, the European Union has adopted several directives that enable member

states to exchange information within the EU on direct and indirect tax matters.15 The

joint OECD/Council of Europe Multilateral Convention on Mutual Administrative

Assistance in Tax Matters, which has been ratified by 8 countries (Denmark, Finland,

Iceland, the Netherlands, Norway, Poland, Sweden, United States), also permits countries

to exchange information on direct and indirect tax matters. Finally, the Nordic

Convention on Mutual Administrative Assistance in Tax Matters allows the Nordic

countries to exchange bank and other information for all kinds of taxes except import

duties. Unlike the other multilateral agreements, the Nordic Convention calls for the

automatic exchange of bank information.

In its survey, the OECD found that 11 members (Australia, Canada, Denmark,

Finland, France, Japan, New Zealand, Norway, Sweden, United Kingdom, the United

States) provided bank information automatically to (some) treaty partners. We requested

these countries to provide additional information about their recipient countries and the

history of this automatic information exchange. The resulting data about the history of

bilateral information exchange are also summarized in Table 5. Several countries

(Australia, Finland, and Norway) mentioned their treaty partners as recipients, but more

generally countries supply information automatically to a more selective and changing

list of countries. The OECD report mentions that Australia, Canada, Denmark, France,

Group on Effective Exchange of Information (including several OECD members and Aruba, Bermuda, Bahrain, Cayman Islands, Cyprus, the Isle of Man, Malta, Mauritius, the Netherlands Antilles, the Seychelles and San Marino). 14 Countries that agree to exchange information automatically typically do not write this into their bilateral tax treaty, but instead conclude a separate memorandum of understanding. 15 In particular, see directives 77/799/EEC, 79/1070/EEC, 92/12/EEC, and the recent proposal COM(2001)294.

12

Norway and Sweden exchange bank information automatically based in part on

reciprocity. As recipient lists of countries vary from year to year and institutional

memories are short, it is impossible to construct an accurate history of bilateral automatic

information exchange.

On the basis of survey responses, however, one can get a relatively complete

picture of automatic information exchange in the BIS-area for 1999 (see Table 6). From

the table, we can see to what extent information exchange in practice occurs on the basis

of reciprocity. Specifically, in the table there are 288 unidirectional entries for which we

also know whether information flows in the other direction. Of these, 67 entries signal the

presence of international information exchange. Of these 67 entries, 30 one-way

exchanges are reciprocated (i.e. there are 15 pairs of bilateral information exchange). To

measure the degree of reciprocity, we constructed 2 dummy variables for our 288

observations flagging whether information was sent and received. The correlation

coefficient between these two dummy variables is found to be 0.28 and to be significant

at the one percent level. This is evidence of reciprocity of information exchange.

A separate issue is whether information exchange and withholding taxes are

complements or substitutes. To investigate this, we note that there are 440 entries for

which we know whether there is information exchange as well as the relevant

withholding tax rate. Breaking down these 440 entries, we find there are 68 entries with

only information exchange, 51 entries with only a withholding tax, 17 entries with both,

and finally 304 entries with neither. The 17 entries with joint information exchange and

withholding taxation all pertain to Australia (as a bank country). Apart from Australia,

information exchange and withholding taxes thus are substitutes rather than

complements.

2.4 Other data

The empirical work below combines the bank liability and tax policy variables

with various controls at the level of the individual country and of the bilateral relationship

between any two countries. National controls are real GDP, the bank interest spread

(defined as the ratio of the lending and deposit interest rates), an index of the rule of law,

and indices of legal system origin. Controls at the bilateral level are trade flows (in both

13

directions), the distance between the two countries, and an index of geographical

contiguity and of a common language. Variables of this type regularly appear in gravity-

type regressions explaining trade or financial flows. Summary statistics of all the data

used in this study are provided in Table 7. The data in the table are for 1999 given that

this year is common to all estimation below. The tax variables in the table, as in the

empirical work, are in the form of tax burdens expressed as percentages of the principal.

Variable definitions and data sources are provided in Appendix A.

3 . Empirical results

This section examines the empirical relationship between tax policy and the

external liabilities of the banking system. As our main interest is in tax policy at the

personal level, we mainly consider non-bank liabilities and deposits. Following Alworth

and Andresen (1992), we use BIS data on bilateral external liabilities and deposits.

Bilateral data are preferred as this allows us to include tax and other information

concerning the bank country, the customer country and their bilateral relationship. The

analysis starts from the following estimating equation:

ijtijtijjtjitiijt XXXI εβββα ++++=

where Iijt is the dependent variable denoting funds held in country i’s banks by residents

of country j (e.g., non-bank external liabilities or non-bank external deposits); next, Xit

are bank country variables (e.g., real GDP), Xjt are customer country variables (e.g., the

wealth tax), and Xijt are characteristics of the bilateral relationship between the bank and

the customer countries (e.g., distance). The vector Xit only contains non-tax-policy

controls, while the vectors Xjt and Xijt contain tax policy variables as well as controls.

Further, α is a constant, the β's are vectors of coefficients, and εijt is an error term. All

regressions in addition include time dummies, while some regressions also contain

country dummies for bank and customer countries alike.

The variable Iijt reflects an equilibrium value in a country’s external liability

market. We consider that banks can operate freely in the international interbank deposit

market, and can obtain funds inelastically at an exogenously given international interbank

14

rate. One reason for this is that the banks themselves generally are not subject to interest

withholding taxes. Changes in tax policy then affect Iijt through their effect on the risk

and after-tax return associated with deposits in different geographical locations as

perceived by international depositors.

Table 8 reports regressions of several measures of external non-bank exposures.

First, regressions of non-bank external liabilities for the period 1983-1999 are reported in

columns (1)-(2). The dependent variable in columns (3)-(4) is non-bank external deposits

for the period 1996-1999, while it is share of non-bank deposits owned by the a country’s

residents held abroad, or s, in columns (5)-(6) again for the period 1996-1999. The

dependent variables as well as the control variables real GDP, bank interest spread and

the two trade variables are in logs. Regressions (2), (4), and (6) include unreported bank

and customer coutry dummies. Regressions (1), (3) and (5) instead include the rule of law

and a set of dummy variables denoting the origin of a country’s legal system as controls.

The bank interest spread serves as a measure of banking system efficiency. Systems with

low interest spreads are expected to be attractive to bank customers and vice versa.

Several estimated coefficients on the bank interest spread variable in the table are

statistically significant and consistent with this. The legal system variables are included

following research by La Porta et al. (1997) showing that the outside equity and debt

finance raised by firms depend importantly on the legal system. The included legal

system variables in Table 8 denote legal systems of French, German and Scandinavian

civil law origins – as opposed to the systems in the English common law tradition. The

generally negative coefficients for these variables suggest that countries with non-English

legal traditions participate less in international bank depositing. More intense

international trade, a smaller distance, geographical contiguity and a common language

are expected to contribute to external bank liabilities. The estimated coefficients in the

table largely confirm these expectations.

Turning to tax policy, the income tax x deposit rate variable is constructed as the

customer-country income tax rate times its deposit interest rate (on the assumption that an

individual depositing in his home country chooses the home currency). This tax variable

obtains positive coefficients and significant coefficients in columns (1) and (2) (be it only

at the 90 percent level in column (2)), but fails to be significant in other regressions. The

15

coefficient of 0.024 in column (2) suggests that a 1 percent increase in the interest tax

burden increases external bank liabilities by 2.4 percent. Next, the wealth tax variable

simply is the wealth tax rate. This variable enters columns (3) a positive and significant

coefficient but it is insignificant in the other columns. The final indicator of customer-

country tax policy is the domestic information variable. This is a dummy variable

flagging the existence of automatic interest information provisioning to domestic tax

authorities. This variable enters columns (1) and (2) with positive and significant

coefficients. The estimated coefficient of 0.248 in column (2) suggests that such domestic

information provisioning increases external bank placements by 28 percent. As indicated

at the bottom of the table, the estimation in columns (1) and (2) includes 7 episodes

where a country adopts a domestic information requirement. During the 1996-1999

period underlying the regressions in columns (3)-(6) no such episode occurred.16

Next, we turn to bank-country tax policy. Withholding tax x deposit rate is

constructed as the non-resident interest withholding tax levied by the bank country times

this country’s deposit interest rate. This variable thus measures the withholding tax

burden the international bank customer faces in the bank country. The withholding tax

variable enters most regressions in the table with negative coefficients, but only

significantly in columns (1), (3), and (5). This reflects that the inclusion of country

dummies suffices to render the coefficient on the withholding tax variable insignificant.

This may reflect that most of the variation in the withholding tax rate is across bank

country.17

A key result in Table 8 is that the interest income tax variable has a significantly

positive impact on external liabilities for the 1983-1999 period, but not on external

deposits for the 1996-1999 period. Most external liabilities in fact are external deposits,

and hence the difference in the results appears to reflect the different time periods. To

16 This explains why a coefficient for the domestic information variable cannot be estimated in regressions (4) and (6) where full sets of country dummies are included. 17 The non-resident interest withholding tax presumably affects a saver’s choice of foreign bank location as much as the more fundamental choice of whether to bank abroad at all. Hence, estimated coefficients on the withholding tax variable may mostly reflect savers’ substitutions among various international banking destinations. Regressions with bilateral data thus cannot tell us directly how aggregate foreign banking would respond if all countries were to raise their withholding taxes (or alternatively were to exchange information).

16

further investigate this, we estimate a regression based on column (2) in Table 8

including the four policy variables interacted with a time dummy for the 1992-1999

period. The non-interacted income tax policy variable enters with a positive and

statistically significant coefficient, while the interacted income tax variable has a negative

coefficient that is statistically significant. Together these results suggest that interest

income taxes mattered in the earlier period of 1983-1991, while the sensitivity of external

deposits to interest income taxes declined from the earlier to the latter period (in fact, the

relationship between external deposits and the income tax variable is statistically

insignificant in the 1992-1999 period).

This is surprising, as reduced transportation and communication costs have

generally increased international capital mobility. One reason why we fail to find a

significant relationship between external deposits and interest income taxes from 1992

may be that the interest income tax burden itself has become almost insignificant – due to

declines in statutory tax rates as well as deposit interest rates (see Figure 1). These

declines probably were motivated by a perceived sensitivity of external liabilities to taxes

in the 1980s, but policy makers may have ‘overshot’ to the point where the tax sensitivity

is no longer material. Another possibility is that the relative importance of individual tax

evaders, as holders of non-bank external liabilities, has declined. Holders of non-bank

deposits that would presumably not respond to personal income tax changes are

corporations, governments, various non-bank tax-exempt financial institutions (such as

mutual funds and insurance companies) and individuals interested in keeping funds

abroad for a variety of non-tax reasons.

Deposit owners in practice may need considerable time to adjust the geographical

location of their deposits to policy changes. To see whether lagged responses are

significant, we report a regression including lagged values for the four policy variables in

column (2) of Table 9. These lagged policy variables fail to be statistically significant.

Substituting the lagged values for the contemporaneous ones – as in column 3 – also

produces lagged policy variables that are statistically insignificant. Hence, there is no

evidence that depositor response to policy changes is stretched out over more than a year.

The income, wealth and non-resident withholding taxes considered in this paper

apply to the interest receipts of individuals. Thus we naturally have considered how tax

17

policy changes affect non-bank external liabilities. The question arises, however, whether

banks adjust their other external lending and borrowing following, say, an increased

inflow of non-bank external liabilities. The two main options are that banks hold more

funds on deposit with international banks or, alternatively, obtain fewer funds on deposit

from international banks. The latter possibility suggests that external bank and non-bank

deposits can be substitutes. To test this, we estimate a regression based on column (2) of

Table 8 with external bank liabilities as the dependent variable. This leads to a coefficient

on the wealth tax variable that is negative and statistically significant, as seen in column

(1) of Table 10. This is in line with the substitution hypothesis, as a higher wealth tax that

encourages external non-bank deposits should discourage external bank deposits. Along

similar lines, the ratio of non-bank to bank liabilities is expected to increase with those

tax policy variables that encourage non-bank external depositing per se. A regression of

this ratio, reported in column (2) of the table, yields positive and significant coefficients

for all three customer country variables, i.e. the interest income tax variable, the wealth

tax variable, and the domestic information variable.

As discussed before, we have been able to collect information on the extent of

bilateral international information exchange only for 1999. To test whether this

information exchange affects external liability flows, we estimate several regressions

with data only for 1999 as reported in Table 11. The first two columns in the table are for

non-bank liabilities, while the remaining two are for non-bank deposits. The inclusion of

full sets of bank and customer country dummies implies that only the bilateral policy

variables (the withholding tax variable and the international information variable) and a

set of bilateral controls can be included. As bilateral trade data tends to be available with

a considerable lag, we find that including the two international trade variables leads to

rather limited samples with 1999 data (as seen in columns (1) and (3)). Therefore, we

also report regressions without these trade variables in columns (2) and (4). The two

policy variables fail to be statistically significant in any of the regressions reported in

Table 11.

The withholding tax variable may by insignificant, as most countries have

adopted zero withholding taxes by 1999. The international information variable further

may not prove to be significant if the exchange of information, as currently organized,

18

fails to bring about an effective tax enforcement. At the same time, international

information cannot have a noticeable effect, if savers by 1999 do not recognize that tax

authorities sometimes ‘automatically’ swap information about particular international

interest payments. Further, international information exchange was far from

comprehensive in 1999 so that savers continued to have access to ‘trusted’ foreign

banking systems with strong reputations for bank secrecy. Continued access to this type

of foreign banking could make information exchange by any subset of countries

ineffectual.

4. Conclusion

This paper has investigated the impact of tax policy on international depositing.

The empirical results indicate that non-bank external liabilities have been positively

related to interest income taxes and to the presence of domestic bank interest reporting.

This is evidence that international deposits are in part intended to facilitate tax evasion.

The sensitivity of international deposits to interest income taxes appears to have declined

after the 1980s. This may reflect that the interest income tax burden itself has been

reduced considerably over the last 2 decades. The financial wealth tax and the non-

resident interest withholding tax burden have similarly been diminished substantially.

As interest withholding taxes have been reduced or eliminated, the international

exchange of information becomes potentially more important to ensure a reasonable

taxation of international interest flows. A simple count of bilateral international

relationships reveals that by 1999 the automatic exchange of information is already as

prominent as withholding taxes. However, we fail to find a significant impact of

international information exchange on international depositing patterns. This justifies

doubts about the effectiveness of international information exchange at present. For the

instrument to become more effective, the quality of the information exchanged may need

to be improved, for instance through the adoption of a common protocol regarding tax

identification numbers. Also, the international exchange of information has to cover most

industrialized countries and other financial centers to be truly effective. All this implies

that international cooperation in this area is necessary to shore up the taxation of

international interest flows.

19

References

Ahearne, Alan G., William L. Griever, and Francis E. Warnock (2000), Informational costs and home bias: an analysis of U.S. holdings of foreign equities, International Finance Discussion Paper 691, Board of Governors of the Federal Reserve System, Washington D.C. Alworth, Julian S. and S. Andresen (1992), The determinants of cross border non-bank deposits and the competitiveness of financial market centres, Money Affairs 5, 105-133.

Bacchetta, Philippe and Maria Paz Espinosa (1995), Information sharing and tax competition among governments, Journal of International Economics 39, 102-21.

Bacchetta, Philippe and Maria Paz Espinosa (2000), Exchange-of-information clauses in international tax treaties, International Tax and Public Finance 7, 275-93.

Bank for International Settlements (2000a), International banking and financial market developments, BIS Quarterly Review, Basel, Switzerland. Bank for International Settlements (2000b), Guide to the international banking statistics, Basel, Switzerland, July 2000. Eijffinger, Sylvester, Harry Huizinga and Jan Lemmen (1998), Short-term and long-term government debt and non-resident withholding taxes, Journal of Public Economics 68, 309-334.

European Commission (1998), Proposal for a council directive to ensure a minimum of effective taxation of savings income in the form of interest payments within the Community, O.J.E.C., COM(1995) 295 final.

European Commission (2001), Proposal for a Council Directive to ensure effective taxation of savings income in the form of interest payments within the Community, COM(2001)400.

Fornari, Fabio and Aviram Levy (2000), Global liquidity in the 1990s: geographical allocation and long-run determinants, mimeo, Bank of Italy.

Grilli, Vittorio. (1989), Europe 1992: issues and prospects for the financial markets, Economic Policy, October. Huizinga, Harry and Søren Bo Nielsen (2000), The taxation of interest in Europe: a minimum withholding tax, in Sijbren Cnossen (ed.): Taxing capital income in the European Union, Oxford University Press.

20

Huizinga, Harry and Søren Bo Nielsen (2003), Withholding taxes or information exchange: the taxation of international interest flows, Journal of Public Economics 87, 39-72. Janeba, Eckhard and Wolfgang Peters (1999), The evasion, tax competition and the gains from non-discrimination: the case of interest taxation in Europe, Economic Journal 109, 93-101.

La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, Robert W. Vishny (1997), Legal determinants of external finance, Journal of Finance 52, 1131-1150.

OECD (2000), Improving access to bank information for tax purposes, Paris. OECD (2002), Agreement on exchange of information on tax matters, Paris.

Phensel (2000), http://garnet.acns.fsu.edu/~phensel/intldata.html#geog.

Portes, Richard, and Hélène Rey (1999), The determinants of cross-border equity flows, CEPR Discussion Paper 2225. WorldAtlas.com (2001), http://www.worldatlas.com.

Zee, Howell H. (1998), Taxation of financial capital in a globalized environment: the role of withholding taxes, National Tax Journal 51, 587-99.

21

Appendix A. Variable definitions and data sources.

External bank liability and deposit data Data on external liabilities and deposits are for all currencies. In the regressions, non-bank external liabilities and non-bank external deposits are in real ecus or euros and in logs. Other dependent variables in Tables 8 and 10 are in logs as well. Total deposits in the banking system in Table 2 are the sum of demand and other deposits (lines 24 and 25 of the International financial statistics of the IMF). Taxation and bank information variables • Income tax x deposit rate = income tax rate (between 0 and 100) times the deposit

interest rate (between 0 and 1) in the bank customer country. The income tax is the final tax paid by residents (either the final withholding tax or the top marginal rate of the personal income tax).

• Wealth tax = wealth tax is the wealth tax applicable to financial assets (between 0 and 100)

• Withholding tax x deposit rate = non-resident interest withholding tax on interest (between 0 and 100) times the deposit interest rate (between 0 and 1) in the bank country

• Domestic information = dummy signaling automatic reporting by banks of interest payments to domestic residents

• International information = dummy signaling the international exchange of information on bank interest payments

The taxation variables are from various issues of International tax summaries (Coopers & Lybrand), International corporate income taxes, a worldwide summary (PriceWaterhouseCoopers), and the European tax handbook (International Bureau for Fiscal Documentation). Information on whether there recently has been domestic interest reporting by banks and any automatic exchange of information on international bank interest payments is taken from OECD (2000). Information on when automatic domestic reporting by banks started and to what countries and since when bank interest information is provided automatically (in Table 5) has been obtained from national authorities. The deposit interest rate is line 60l of the International financial statistics of the IMF. Other variables • Real GDP = log of GDP in real ecus or euros • Bank interest spread = ratio of bank lending and deposit interest rates (in logs in

regressions) • Rule of law = assessment of law and order in a country. Average of the months of

April and October of the monthly index between 1982 and 1995. On a scale from 0 to 10 with lower scores for less law and order. The variable is an assessment of the strength and impartiality of the legal system and of popular observance of the law (see La Porta et al., 1997)

22

• French law = dummy identifying French legal origin • German law = dummy identifying German legal origin • Scandinavian law = dummy identifying Scandinavian legal origin • Bank country exports = exports from bank country to customer country in real ecus

or euros (in logs in regressions) • Customer country exports = exports from customer country to bank country in real

ecus or euros (in logs in regressions) • Distance = distance in kilometers from capital to capital (in logs in regressions) • Contiguity = dummy identifying a common border. • Common language = dummy identifying if a pair of countries has at least one

common language. Data on GDPs and trade are from Eurostat and the IMF. The lending interest rates if from line 60p of the International financial statistics of the IMF. Information on rule of law and legal origin is from La Porta et al. (1997). Data on distance, contiguity, and common language are from WorldAtlas.com (2000) and Phensel (2000).

23

Table 1. External liabilities and deposits of banks in the BIS-area in 1999

External liabilities External deposits

� bn % non-bank � bn % non-bank

Australia 146 8 47 26

Austria 80 12 65 15

Bahamas 225 33 224 33

Bahrain 82 31 82 31

Belgium 272 31 261 28

Canada 100 32 95 34

Cayman Islands 604 42 597 43

Denmark 56 15 46 18

Finland 22 20 12 35

France 611 9 472 12

Germany 819 32 719 37

Hong Kong 349 23 348 23

Ireland 129 19 126 19

Italy 233 7 232 7

Japan 509 6 502 6

Luxembourg 371 37 319 37

Netherlands 288 18 240 22

Norway 25 9 15 12

Portugal 65 17 55 13

Singapore 393 29 361 32

Spain 184 39 177 40

Sweden 72 13 53 10

Switzerland 560 48 560 48

United Kingdom 1,778 21 1,626 21

United States 1,035 9 1,035 13

Other 24 30 24 30

Total 9,031 24 8,292 25

Source: BIS (2000), Tables 2A, 2B, 3A, and 3B and own calculations

24

Table 2. Summary statistics on external deposits in 1998

Country Exports of deposits (� bn)

Imports of deposits (� bn)

Exports of non-bank deposits (� bn)

Imports of non-bank deposits (� bn)

Total non-bank deposits

in banking system (� bn)

Non-bank deposits

owned by residents held

at home or abroad (� bn)

Share of non-bank deposits

owned by residents held

abroad (%)

Expansion ratio of non-bank deposits in

banking system due to net

imports of non-bank deposits

(%) Australia 16 18 6 3 203 207 3 -2 Austria 37 43 6 6

Bahamas 124 147 12 15 Belgium 154 185 13 30 Canada 45 75 14 15 317 316 4 0

Denmark 33 36 3 5 86 84 3 3 Finland 17 9 1 1 53 53 2 -1 France 288 326 41 34

Germany 337 494 87 97 1,267 1,257 7 1 Ireland 58 91 19 19 57 57 33 0 Italy 155 173 41 17 457 481 9 -5 Japan 348 364 36 14

Netherlands 254 202 Norway 6 10 2 1 72 73 3 -1 Portugal 29 30 5 5 87 87 6 0

Spain 112 108 46 18 317 346 13 -8 Sweden 31 57 4 12

Switzerland 459 261 42 93 325 273 15 19 United Kingdom 1,035 1,024 86 237

United States 656 541 228 31 2,291 2,488 9 -8 For data sources see Appendix A. Note that exports and imports are calculated using only data from those countries for which imports are available.

25

Table 3. Summary statistics on external liabilities in 1998

Country Exports of liabilities

(� bn)

Imports of liabilities

(� bn)

Exports of non-bank liabilities

(� bn)

Imports of non-bank liabilities

(� bn) Australia 22 74 7 3 Austria 38 45 6 6

Bahamas 129 157 10 59 Bahrain 25 35 2 3 Belgium 157 200 13 41 Canada 48 81 15 18

Denmark 38 37 3 5 Finland 17 9 1 1 France 313 348 41 35

Germany 352 512 89 97 Hong Kong 258 294 23 14

Ireland 59 92 19 19 Italy 158 177 41 17 Japan 653 545 40 15

Netherlands 268 209 Norway 7 10 2 1 Portugal 29 40 5 8

Singapore 216 251 Spain 113 109 47 18

Sweden 32 58 5 12 Switzerland 463 277 42 98

United Kingdom 1,129 1,098 97 243 United States 709 572 234 34

For data sources see Appendix A. Note that exports and imports are calculated using only data from those countries for which imports are available.

26

Table 4. Wealth tax, interest income tax, and non-resident withholding tax for bank deposits in 1999.

Country Income tax18 Wealth tax Withholding tax for non-residents Australia 47 0 10 Austria 25 0 0 Bahamas 0 0 0 Bahrain 0 0 0 Belgium 15 0 0 Canada19 48.75 0 0 Cayman Islands 0 0 0 Denmark20 61.7 0 0 Finland21 28 0.9 0 France 22 25 1.8 0 Germany23 56.975 0 0 Hong Kong 0 0 0 Ireland 24 0 0 Italy 27 0 0 Japan24 20 0 10/15 Luxembourg 47.15 0.5 0 Netherlands 60 0.7 0 Netherlands Antilles 60 0 0 Norway25 28 1.1 0 Portugal 20 0 10/12/15/20 Singapore 28 0 0 Spain26 48 2.5 0 Sweden 30 1.5 0 Switzerland27 41.4 0.713 0/5/10/12.5/35 United Kingdom 40 0 0 United States28 39.6 0 0

For data sources see Appendix A

18 Final withholding tax or top marginal tax rate. 19 Ontario. 20 Copenhagen. Sum of basic rate, surcharges, and local and church taxes. 21 Helsinki. 22 Including social surcharge and generalized social tax. 23 Including solidarity surcharge. 24 Tokyo. Including local taxes. 25 Sum of 0.4 percent national tax plus 0.7 percent local tax. 26 Including regional tax. 27 Bern, including cantonal and municipal wealth tax. 28 Federal tax only.

27

Table 5. Domestic and international reporting of bank interest payments

Automatic reporting by banks on interest payments to

domestic residents

International automatic exchange of information on bank interest payments

Country Yes or no If yes, since To Since Australia Yes 88 Treaty partners About 95 Austria No None Belgium No None Canada Yes U.S at least Denmark29 Yes 77 Differing countries 1993

Finland30 Yes Over 20 years Treaty partners (except Russia)

Over 20 years

France31 Yes 84 94 Germany No None Greece No None Ireland Yes 92 None Italy No None Japan Yes Some countries Luxembourg No None Netherlands Yes 87 None Norway Yes 86 Treaty partners More than 10

years Portugal No None Spain Yes 85 None Sweden Yes 86 Canada, Denmark,

Finland, Iceland, Norway, US.

Australia, Estonia,

France, Italy, Japan, Lithuania, Spain, UK.

91

97

Switzerland No None United Kingdom Yes 52 Some countries United States Yes Canada 1997

For data sources see Appendix A

29 In 1998 and 1999, Denmark provided info to Australia, Canada, Czech Republic, Faeroe Islands, Finland, France, Greenland, Hungary, Japan, Korea, New Zealand, Norway, Spain, Sweden, UK, US. 30 Main recipients have been Belgium, Denmark, Finland, France, Germany, Iceland, Japan, New Zealand, Poland, Sweden, UK, US. 31 In 1999, France provided information to Australia, Austria, Belgium, Canada, Denmark, Finland, Germany, Iceland, Italy, Japan, Korea, Netherlands, New Caledonia, New Zealand, Norway, Portugal, Spain, Sweden, UK, US.

28

Table 6. International automatic exchange of information on bank interest in 1999

Receiving Country

\

Providing Country

Aus

tral

ia

Aus

tria

Bah

amas

Bah

rain

Bel

gium

Can

ada

Cay

man

Isl.

Den

mar

k

Finl

and

Fran

ce

Ger

man

y

Gre

ece

Hon

g K

ong

Irel

and

Ital

y

Japa

n

Lux

embo

urg

Net

herl

ands

Net

herl

. Ant

.

Nor

way

Port

ugal

Spai

n

Swed

en

Switz

erla

nd

U. K

ingd

om

Uni

ted

Stat

es

Australia X 1 0 0 1 1 0 1 1 1 1 0 0 1 1 1 0 1 0 1 0 1 1 1 1 1 Austria 0 X 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Belgium 0 0 0 0 X 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Canada X 1

Denmark 1 0 0 0 0 1 0 X 1 1 0 0 0 0 0 1 0 0 0 1 0 1 1 0 1 1 Finland 1 1 0 0 1 1 0 1 X 1 1 1 0 1 1 1 1 1 0 1 1 1 1 1 1 1 France 1 1 0 0 1 1 0 1 1 X 1 0 0 0 1 1 0 1 0 1 1 1 1 0 1 1

Germany 0 0 0 0 0 0 0 0 0 0 X 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Greece 0 0 0 0 0 0 0 0 0 0 0 x 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Ireland 0 0 0 0 0 0 0 0 0 0 0 0 0 X 0 0 0 0 0 0 0 0 0 0 0 0

Italy 0 0 0 0 0 0 0 0 0 0 0 0 0 0 X 0 0 0 0 0 0 0 0 0 0 0 Luxembourg 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 X 0 0 0 0 0 0 0 0 0 Netherlands 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 X 0 0 0 0 0 0 0 0

Norway 0 0 1 0 1 1 1 1 0 1 0 X 1 1 1 Portugal 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 X 0 0 0 0 0

Spain 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 X 0 0 0 0 Sweden 1 0 0 0 0 1 0 1 1 1 0 0 0 0 1 1 0 0 0 1 0 1 X 0 1 1

Switzerland 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 X 0 0 United States 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 X For data sources see Appendix A.

29

Table 7. Summary statistics for 1999.

Variable #obs Unit Mean Std. deviation Minimum Maximum

Country variables

GDP 23 � Bn 975.862 1873.174 6.212 8638.711

Bank interest spread

22 4.628 7.326 1.261 33.43332

Rule of law 20 9.228 1.066 6.180 10

French law 22 0.318 0.477 0 1

German law 22 0.182 0.395 0 1

Scandinavian law

22 0.182 0.395 0 1

Income tax x deposit rate*

23 0.938 0.874 0 3.166

Wealth tax 26 0.374 0.675 0 2.500

Domestic information

20 0.650 0.489 0 1

Bilateral variables

Bank country exports

286 � Bn 5.055 15.479 0.000 192.366

Customer country exports

286 � Bn 4.275 11.216 0.000 135.511

Distance 702 Km 5942 4760 174 18,389

Contiguity 702 0.071 0.257 0 1

Language 702 0.185 0.389 0 1

Withholding tax x deposit

rate*

575 0.040 0.115 0 0.480

International information

440 0.193 .0395 0 1

Non-bank liabilities

520 � Bn 2.235 7.769 0.000 90.567

Bank liabilities 520 � Bn 8.669 23.155 0.000 182.108

Non-bank deposits

468 � Bn 2.289 8.048 0.000 89.833

Bank deposits 468 � Bn 8.860 23.196 0.000 182.108 * as percentage of principal. For data sources, see Appendix A

32 Ireland’s deposit rate was 0.10 percent and its loan rate was 3.34 percent in 1999.

30

Table 8. Determinants of external non-bank liabilities and deposits Non-bank liabilities Non-bank deposits Non-bank deposits

divided by ownership (1) (2) (3) (4) (5) (6)

Bank country Real GDP .082

(.056) .852* (.431)

-.122** (.019)

2.683 (1.952)

.191 (.143)

3.152 (2.299)

Bank interest spread -.386** (.063)

.085 (.065)

-.528** (.100)

-.267 (.240)

-.304* (.138)

-.147 (.335)

Rule of law .667** (.067)

.021 (.104)

.134 (.141)

French law -.249* (.116)

-.776** (.160)

-.927** (.216)

German law -.146 (.128)

.193 (.186)

-.272 (.266)

Scandinavian law -2.719** (.138)

-2.309** (.193)

-2.729** (.249)

Customer country Real GDP .184**

(.058) .890

(.492) .081

(.104) .056

(1.861) -.629** (.193)

1.141 (3.209)

Bank interest spread -.094 (.070)

.250** (.073)

.039 (.152)

-.292 (.272)

-.386 (.428)

-.212 (.774)

Rule of law .139** (.045)

.085 (.105)

-.311 (.247)

French law -.988** (.097)

-1.233** (.203)

-2.172** (.389)

German law -.184 (.142)

-.573 (.324)

-1.483 (.959)

Scandinavian law -2.285** (.116)

-2.469** (.210)

-2.526** (.305)

Income tax x deposit rate .056** (.017)

.024 (.014)

.018 (.098)

-.074 (.141)

.006 (.172)

-.125 (.214)

Wealth tax .064 (.047)

.065 (.068)

.212* (.105)

.164 (.324)

.175 (.285)

.124 (.366)

Domestic information .367** (.091)

.248* (.124)

.190 (.215)

-.041 (.580)

Relationship Bank country exports .349**

(.057) .253** (.056)

.540** (.101)

.443** (.100)

.065 (.144)

.500 (.399)

Customer country exports .239** (.058)

.388** (.047)

.246* (.106)

.439** (.085)

.420** (.147)

.395** (.116)

Distance -.838** (.060)

-.234** (.079)

-.741** (.103)

-.140 (.131)

-1.396** (.166)

-.427* (.207)

Contiguity .000 (.094)

.391** (.086)

-.119 (.166)

.249 (.154)

-.401 (.223)

-.197 (.215)

Common language .442** (.109)

.529** (.078)

.219 (.182)

.058 (.130)

.336 (.229)

.049 (.155)

Withholding tax x deposit rate

-.781** (.112)

.022 (.106)

-1.414** (.249)

-.486 (.368)

-1.226** (.331)

-.401 (.581)

Adj. R² .72 .84 .70 .83 .70 .83 No. of obs 2375 2375 757 757 410 410 No. changes in domestic information

7 7 0 0 0 0

Data on liabilities are for 1983-1999, while data on deposits is for 1996-1999. All regressions include unreported time dummies. Columns (2), (4) and (6) include bank and customer country dummies. The sample underlying columns (5) and (6) only includes data for those 18 countries for which we can compute both non-bank deposit exports and imports during the 1996-1999 period. Detailed variable definitions and data sources are given in Appendix A. Heteroskedasticity consistent errors are given in parentheses. * and ** indicate significance levels of 5 and 1 percent, respectively.

31

Table 9. Determinants of external non-bank liabilities for different time periods

(1) (2) (3) Bank country non-policy variables

Real GDP .880* (.429)

1.126* (.440)

1.168** (.439)

Bank interest spread .087 (.065)

.082 (.070)

.074 (.070)

Customer country non-policy variables

Real GDP 1.243* (.532)

.603 (.547)

.879 (.533)

Bank interest spread .188* (.014)

.252** (.079)

.554** (.078)

Relationship non-policy variables

Bank country exports .247** (.056)

.246** (.060)

.257** (.060)

Customer country exports .386** (.047)

.405** (.050)

.398** (.050)

Distance -.242** (.079)

-.249** (.083)

-.238** (.084)

Contiguity .393** (.085)

.367** (.090)

.370** (.090)

Common language .533** (.078)

.508** (.083)

.518** (.083)

Policy variables

Income tax x deposit rate .035* (.014)

.047 (.027)

Wealth tax .035 (.069)

-.007 (.116)

Domestic information .250* (.126)

.392* (.176)

Withholding tax x deposit rate -.029 (.111)

.001 (.195)

Income tax x deposit rate, 1992-1999 -.101* (.046)

Wealth tax, 1992-1999 .069 (.058)

Domestic information, 1992-1999 -.125 (.110)

Withholding tax x deposit rate, 1992-1999 .148 (.131)

Income tax x deposit rate-1 -.026 (.024)

.013 (.013)

Wealth tax-1 .108 (.115)

.091 (.071)

Domestic information-1 -.089 (.151)

.130 (.128)

Withholding tax x deposit rate-1 .118 (.180)

.108 (.112)

Adj. R² .84 .84 .84

No. of obs 2375 2213 2216

No. changes in domestic information 7 6 6

Data on liabilities are for 1983-1999. All regressions include unreported time and bank and customer country dummies. Detailed variable definitions and data sources are given in Appendix A. Heteroskedasticity consistent errors are given in parentheses. * and ** indicate significance levels of 5 and 1 percent respectively.

32

Table 10. Determinants of external bank liabilities

Bank liabilities Ratio of non-bank to bank liabilities

(1) (2) Bank country

Real GDP 3.191** (.360)

-2.277** (.471)

Bank interest spread .249** (.063)

-.155* (.072)

Customer country

Real GDP 1.999** (.466)

-1.217* (.562)

Bank interest spread .272** (.063)

-.007 (.081)

Income tax x and deposit rate -.028 (.015)

.052** (.016)

Wealth tax -.207** (.057)

.279** (.074)

Domestic information -.118 (.108)

.264* (.126)

Relationship

Bank country exports .603** (.049)

-.334** (.059)

Customer country exports .363** (.050)

.024 (.054)

Distance -.036 (.070)

-.198* (.088)

Contiguity -.280** (.076)

.668** (.095)

Common language .306** (.080)

.184 (.095)

Withholding tax x deposit rate .032 (.095)

-.016 (.120)

Adj. R² .87 .61 No. of obs 2465 2371 No. changes in domestic information

7 7

Data on liabilities is for 1983-1999. All regressions include unreported time dummies and bank and customer country dummies. Detailed variable definitions and data sources are given in Appendix A. Heteroskedasticity consistent errors are given in parentheses. * and ** indicate significance levels of 5 and 1 percent, respectively.

33

Table 11. Determinants of non-bank liabilities and deposits in 1999 Non-bank liabilities Non-bank

deposits

(1) (2) (3) (4) Exports .298

(.220) .283

(.697)

Imports .429 (.222)

.468* (.221)

Distance -.738 (.422)

-1.437** (.269)

-.718 (.429)

-1.441** (.268)

Contiguity .223 (.372)

.441 (.316)

.221 (.375)

.433 (.315)

Common language -.250 (.277)

-.433 (.308)

-.251 (.273)

-.442 (.301)

Withholding tax x deposit rate

6.384* (3.126)

2.227 (3.008)

6.076 (3.133)

1.914 (3.138)

International information .304 (.380)

.036 (.351)

.336 (.379)

.007 (.345)

Adj. R² .78 .74 .78 .73 No. of obs. 112 203 112 203

All regressions include unreported time dummies as well as unreported bank and customer country dummies. Detailed variable definitions and data sources are given in Appendix A. Heteroskedasticity consistent errors are given in parentheses. * and ** indicate significance levels of 5 and 1 percent, respectively.

34

Figure 1. Average interest income tax on interests payments from domestic deposits to resident individuals

0

10

20

30

40

50

60

1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

per

cent

0

1

2

3

4

5

6

per

cent

average interest income tax rate on residents (left-hand scale) Interest income tax burden (right-hand scale)

Note. Non-weighted average for countries listed in Table 6. For data sources see Appendix A Figure 2. Average wealth tax on financial wealth

0.35

0.40

0.45

0.50

0.55

0.60

0.65

1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

per

cent

Note. Non-weighted average for countries listed in Table 6. For data sources see Appendix A

35

Figure 3. Average withholding tax on interest from bank deposits to non-residents

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

per

cent

.0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

per

cent

Withholding tax rate on interest from bank deposits to non-residents (right-hand scale) withholding tax burden (left-hand scale) Note. Non-weighted average for countries listed in Table 6. The increase in the early 90’s is due to the introduction of a withholding tax in Greece and the extension of the Austrian withholding tax to a larger set of countries. The sharp decrease in 1997 and 1998 is mainly due to changes in the Greek and the Italian withholding tax rates. For data sources see Appendix A.